Predictive Analytics World
Last week I attended the second gathering of Predictive Analytics World, a conference of predictive analytics (PA) practitioners in which they share case studies, best practices, tips, tricks and war stories. The days there proved to be an invaluable resource, and I’ve outlined below just a handful of my key takeaways from the conference:
Our profession’s first challenge lies in describing what we do – answering the question “what is predictive analytics?” is no easy task. Definitions at the conference ranged from the poetic: author Stephen Baker described it as the mathematical modeling of humanity; to the practical: Dr. John Elder, a premier PA consultant, offered that predictive analytics “fills information gaps in difficult business decisions.”
Another more academic concern is distinguishing our nascent industry from our disciplinary cousins, statistics and data-mining. To be clear, they are a component of the industry, but as stand-alone pursuits, they are outshone by PA. Predictive analytics is by definition future-oriented, and it’s outcomes are punctuated with a dollar sign, having direct consequences to a company’s bottom line.
Most importantly, the conference emphasized for me the strategic importance of data within an organization. As data capture, storage and retrieval mechanisms continue to improve, data’s value to a company as a strategic resource increases exponentially. Companies, candidates and campaigns that segregate, limit or otherwise hide their data under a bushel are doomed.
Finally, it was encouraging to hear others concur with our own findings: that in the world of PA real learning occurs over time. Data begets more data, and the longer you’ve been engaged in meaningful analysis of that data, the more you can leverage it to improve its power. Fortunately for us, this is a key advantage of conservatives’ early investment in data (via VoterVault) and microtargeting years before liberals began their own efforts in earnest.
- Alex Lundry
Comments on this entry